project-governance

Warn

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a task contract system to resolve and execute shell commands via subprocess.run. These commands are dynamically constructed from repository-local configuration files like config.yaml and specific task contracts. While the skill requires a --authorized flag for mutating operations, the actual execution path is dependent on the integrity of the project's configuration files. (Files: scripts/project-governance.py, scripts/release-workflow.py)
  • [REMOTE_CODE_EXECUTION]: The release workflow executes project-defined hooks (e.g., artifact.freeze, deploy, verify) which are specified in a release-workflow.json file within the repository being governed. This allows the execution of arbitrary code paths defined outside of the skill's own package. (File: scripts/release-workflow.py)
  • [COMMAND_EXECUTION]: Multiple scripts within the skill (document-maintenance.py, domain-knowledge.py, validate-governance.mjs) attempt to locate and execute an external Python script (mdq.py) from other skill installation directories. This creates a dependency on external code that may vary between environments. (Files: scripts/document-maintenance.py, scripts/domain-knowledge.py, scripts/validate-governance.mjs)
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 22, 2026, 05:31 AM
Security Audit — agent-trust-hub — project-governance